Human Attuned Machine Learning Tool for Dune Annotation in ArcGIS Pro
Abstract
The Human-Attuned Machine Learning (HAML) Dune Annotation Tool (DAT) is a custom ArcGIS Pro plugin that facilitates quick, simple, and accurate annotation of coastal dune formations based on digital elevation models collected from LiDAR or other remote sensing methodologies. It implements an intuitive user interface consisting of basic lines and vertices which are first generated by a machine-learning backend and then adjusted or corrected by the user, if necessary, in an iterative process following the methods of previous HAML work. The intended purpose of the tool is to enable geospatial analysts to efficiently and accurately annotate coastal dune formations, so the tool is designed to follow the standard procedures for detecting dune crests and dune toes that have previously been used to conduct studies on the coastal response to storm surges and other adverse coastal phenomena. It utilizes several fail-safes and quality control checks on the output to minimize the amount of manual user corrections needed. This tool is part of ongoing research into the further optimization of geographic region annotation.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 13, 2023
- Accession Number
- AD1206091
Entities
People
- Benji J Lee
- Bradley Landreneau
- Christopher J. Michael
- Nicholas M. Studer
- Steven M. Dennis
Organizations
- United States Naval Research Laboratory